Samsung Electronics has decided to supply its next-generation high-bandwidth memory (HBM4) chips to OpenAI. These chips will be used in OpenAI’s first in-house artificial intelligence processor, according to The Korean Daily on Wednesday.
Last year, Samsung signed an official letter of intent to supply memory chips for OpenAI’s data centre, aiming to meet the growing demand for its Stargate project. The company plans to provide up to 800 million gigabytes (GB) of 12-layer HBM4 chips to OpenAI by the second half of 2026.
These HBM4 chips will work alongside OpenAI’s first AI processor, a custom chip developed in collaboration with Broadcom. The processor is expected to be produced by Taiwan Semiconductor Manufacturing Company (TSMC) and is expected to start production in the third quarter of the year, with a launch by the end of the year.
Last year, OpenAI partnered up with Broadcom to develop its first-ever in-house AI processors. This latest partnership is another step for the maker of ChatGPT in its efforts to obtain the computing power it needs to meet the growing demand for its services.
On Wednesday, Samsung and Advanced Micro Devices (AMD) said they had a memorandum of understanding (MoU) to further develop their strategic partnership in providing memory chip solutions for artificial intelligence infrastructure. Under this partnership, Samsung will be a key supplier of HBM4 chip solutions for AMD’s upcoming AI GPUs.
According to sources, Samsung Electronics did not comment, and OpenAI was unavailable for comment outside regular business hours.
